calib.fit {calib}R Documentation

General Standard Curve Fitting

Description

calib.fit utilizes two main model types, linear and logistic, for the purpose of standard curve fitting. It also incorporates several alternatives within each model type to allow for more flexible and reliable fitting.

Usage

calib.fit(x, y, m = 2, cv = 0.2, conf = 0.95, b3start, 
b4start, mx = 50, ap.calc = T, lof.calc = T, theta, 
method = "squared", kmax = 10, theta.init = 1, digits = 3, 
type = c("log.fpl.pom", "fpl.pom", "log.fpl", "fpl", 
"lin.quad.pom", "log.thpl.pom", "thpl.pom", "log.thpl", 
"thpl", "lin.pom"), intercept = T)

Arguments

x Independent variable (for example dose)
y Dependent variable (for example response)
m Number of repeated measurements
cv The acceptable coefficient of variation. The limits of quantitation are calculated with this constraint
conf The confidence level used for the determining the prediction interval
b3start The starting value for b3
b4start The starting value for b4
mx The maximum number of iterations used in the non-linear least-squares fit
ap.calc Should assay performance measures (MDC, RDL and LOQ) be calculated. The default is TRUE.
lof.calc Should the lack of fit statistics be calculated. The default is TRUE.
theta If a value is specified for theta then calib.fit will not estimate theta and will only perform the iterative reweighting of the response.
method The loss function used in the optimization process when estimating theta. The values for method can be "absolute", "squared", "ols", or "reml". The default is "squared".
kmax The maximum number of iterations used to estimate theta. The default is 10.
theta.init The initial starting value for theta.
digits A graphing parameter specifying the number of digits which will be displayed when an object of class calib.fit is plotted.
type Can take the values "log.fpl.pom", the log parameterized four parameter logistic regression (fpl) fit with power of the mean (POM), "fpl.pom", fpl fit with POM, "log.fpl", log parameterized fpl, "fpl", standard fpl, "thpl.pom", three parameter logistic (thpl) regression fit with POM, thpl without POM "thpl", log parameterized POM thpl, "log.thpl.pom", log parameterized thpl, "log.thpl", "lin.quad.pom", linear regression with a quadrattic term fit by POM, "lin.pom", linear regression fit by POM.
intercept Whether or not to include an intercept in a model of type "lin.quad.pom" or "lin.pom". Default is TRUE.

Value

coefficients Estimates of the coefficients
se.coefficients Estimates of the standard errors for the coefficients
cov.unscaled Unscaled variance-covariance matrix of the coefficients
theta
df.residuals The residual degree of freedom
fitted.values The estimated fitted values
residuals The values of the residuals
vfe.method The type of variance function used to estimate theta
kused The number of iteration to convergence of the GLS fit
status Indicated whether the algorithm converged
x The x values
y
parm Indicated whether a log parameterization of the model used
m The number of repeated measures used in the model
cv Coefficient of variation used
mdc Minimum detectable concentration
rdl Reliable detection limit
loq Limit of quantitation
cf Measure of assay performance
gradient The gradient matrix based off of the final parameter estimates
Fstat The F statistic for the goodness of fit test
p.value The p-value for the goodness of fit test
lofss Lack of fit sum of squares
df.lof Lack of fit degrees of freedom
pure.error Lack of fit error deviation from the mean
df.pure.error Degrees of freedom for pure error
sse Lack of fit error deviation from the fitted values
sse Degrees of freedom for sse
var.model The type of variance model used
conf.level Confidence level used
mmod In the linear regression fit whether a quadrattic or linear model was fit.
SST Sum of squares total. Linear regression fit only.
SSE Sum of squares error. Linear regression fit only.
SSR Sum of squares regression. Linear regression fit only.
Rsq R squared. Linear regression fit only.
type The general model type used, fpl, thpl or lin.

Author(s)

Perry Haaland, Daniel Samarov, Elaine McVey

References

Davidian and Haaland 1992

Examples

data(HPLC)
attach(HPLC)
model <- calib.fit(Concentration, Response)

[Package calib version 0.1.02 Index]